This project aims to utilize a machine learning method proposed by Ling et al., the tensor basis neural network (TBNN), to learn a model for the Reynolds stress anisotropy tensor of a turbulent channel flow from the DNS data. The data used in the project is obtained from the Johns Hopkins Turbulence Database (JHTDB). See full status report here.
To run the codes in this repository, make sure you have installed the following packages:
- pyJHTDB at https://github.com/idies/pyJHTDB
- tbnn at https://github.com/tbnn/tbnn
Notice that different python versions are required for codes in different folders:
- ./JHTDB -> Python 3
- ./TBNN -> Python 2